How should you interpret the PyStone value across machines?

Question

I have three different laptops with Python2.5 on it. If I run Pystone I get essentially the same values.

• Intel Pentium M @1.80GHz - 1.5GB RAM
• Intel T2300 @1.66GHz - 1GB RAM
• Intel Pentium Dual T2370 @1.73Ghz - 2GB RAM

These are their respective Pystone values:

• 45787 pystones/second
• 44377 pystones/second
• 43861 pystones/second

I don't think less is better but my oldest machine scores higher and the most recent scores the lowest. Why would this happen?

Answer 1

Your CPUs score 442, 806, and 869 respectively on Passmark's list of CPU benchmarks. When running on a single core, your dual-core CPUs should perform roughly equivalent to a CPU with half their score. The single-core scores then are about 442, 403, 434 (i.e., all about the same, give or take a bit for clock/bus speed and cache size).

I assume pystone doesn't take advantage of multiprocessing, and your results back this up.

So, to answer your question, all of your processor cores are approximately the same performance class, so they all do approximately the same on the pystone test. The dual-core systems should be able to run two concurrent pystone's without much performance penalty though!

Here are a few other benchmarks from machines that I have handy (python 2.5 or 2.6):

                          Pystone   Passmark
Dual PowerMac G4 @ 1GHz:   18587        n/a (probably about 300)
Pentium 4 @ 2.53GHz:       37313        327
Core2 Duo T7200 @ 2GHz:    51245       1129

Answer 2

PyStone is not a good or reliable benchmark, so you should probably not interpret them at all. But, no, higher is better, not the other way around. Also be sure to run it multiple times after each other, many CPU's slow down when idle today.